Resource Management In Service Overlay Network

With the development of high-speed networks and communications, streaming media technology, a variety of new-style applications, VoIP, VOD and others with higher requirements in particular, have emerged in the Internet. The emergence of these new-style applications makes Internet's "best effort" service system encounter serious challenges. However, Internet has already been a complex system connected by a large number of heterogeneous autonomous domains, which makes it hard to introduce new capabilities or replace current technologies with something better. So solutions are accordingly found from the application layer, and SON networks come into being.SON is the kind of overlay network built upon existing infrastructure network and operated by a third party ISP. It buys bandwidth with a certain QoS guarantee from the underlying network to build up the virtual data transmission network, and benefits from end-to-end QoS service for end users. Its operation needs good business relationship. In short, SON utilizes covering techniques to provide value-added services in the Internet. But it is undeniable that the deployment of SON is a capital intensive investment. From overlay service providers'point of view, the cost, including bandwidth cost and node cost has to be taken into consideration. To reduce the cost, the optimal bandwidth allocation is currently studied by scholars, while the study on node placement is seldom. In addition, another way to minimize the bandwidth cost is to maximize the use efficiency to obtain greater profits. Therefore, this paper aims to study the placement of nodes and bandwidth utilization and other resource management problems, structured as follows:(1) The existence of SON depends on users who are willing to pay more for higher quality network service. So the precondition of its existence is that users can access SON. On this basis, this paper focuses on how to control the number of nodes and the node location and proposes a node placement model, giving two greedy algorithms. Moreover, the experiment of the model is also conducted to prove its applicability. (2) The appropriate selection of overlay link can effectively guarantee the performance of SON, and it helps control bandwidth costs, after all, the bandwidth price of the underlying network are different. In addition, in order to achieve the purpose of using network resources simultaneously, it is necessary to make the best use of the diversity of overlay path, so when computing the topology of SON, whether multipath can be found between any nodes is also considered.(3) Multipath routing can significantly improve the utilization of network resources such as bandwidth, but it is very difficult to do multipath selection. Because overlay links which are seemingly unrelated are likely to reuse some underlying physical links, especially the bottleneck links, if not carefully selected, the performance of SON is going to be much lower. Therefore, a minimum-related multipath select model is proposed. Since it is an NP-hard problem, this paper then introduces a similar Minimal Early-Path Correlated Paths algorithm MECP and the corresponding multi-resource allocation algorithm. Simulation results show that, MECP algorithm can effectively satisfy QoS users, increase network throughput, and help balance the load and network performance stability.